Products for Internet platforms can be divided into two main categories: goods and services. To improve the sales of products through data analysis, you first need to understand what data needs to be analyzed?
What data needs to be analyzed?
One, the operation module
From the point of view of the user's consumption process, it can be divided into four parts: attraction, conversion, consumption, retention.
Traffic
Traffic is mainly reflected in the attraction link, according to the traffic structure can be divided into channel structure, business structure and regional structure. Channel structure, you can track the traffic situation of each channel and analyze the quality of each channel through the proportion of channel traffic. Business structure, according to the designated business to track the activity of the traffic, to observe the activity before, during and after the change of traffic, to make an assessment of the effect of the activity.
Conversion Rate
Conversion Rate=Number of Desired Behaviors/Total Number of Actions. Increased conversion rate means lower cost and higher profit, the most classic analytical model is the funnel model.
Churn rate and retention rate
Through various channels or activities to attract users, but after a period of time there will be users lost, this part of the user is the loss of users, and stay in this part of the user is the retention of users. Churn can be divided into rigid churn, experience churn and competition churn, although churn is inevitable, but according to the analysis of the churn, make the appropriate countermeasures to retain users. Regarding retention, by observing the pattern of retention, positioning the retention stage, you can assist in marketing activities, marketing strategy positioning, etc., but also can compare the retention of different users, product features, analyze the value of the product, and make timely adjustments to the product.
Repurchase rate
Repurchase rate can be divided into "user repurchase rate" and "order repurchase rate", through the analysis of the repurchase rate, you can further analyze the user's stickiness, to assist in the discovery of the repurchase rate problem, the development of operational strategies, colleagues. You can also do horizontal (product, user, channel) comparison analysis to refine the repurchase rate and assist in problem positioning.
Two, sales module
Sales module has a large number of indicators, including the same ring ratio, completion rate, sales ranking, the proportion of key products, the proportion of platforms and so on.
Three, the commodity module
Important indicators of analysis: including age, sales rate, out-of-stock rate, structural indicators, pricing system, correlation analysis, smooth sales analysis, etc., used to judge the value of the goods, to assist in adjusting the commodity strategy
Four, the user module
Focus on the analysis of the indicators: including the number of new users, the growth rate, the turnover rate, the percentage of effective members, Retention and so on
User value analysis: according to the RFM model, and then incorporate other personalized parameters, the value of the user's division, and for each level of users to make further analysis.
User Profile: Based on inherent attributes, behavioral attributes, transactional attributes, interests and hobbies and other dimensions, we can add labels and weights to users, design user profiles, and provide accurate marketing reference.
Selecting an analysis model based on the data to be analyzed
A user model
User model is a way to depict target users in marketing planning or business design, and it often comes in a variety of combinations, which makes it easy for planners to use it to analyze and set up their strategies for different users. There are two traditional approaches to user modeling: interview and observation-based user modeling (rigorous and reliable, but time-consuming), and ad hoc user modeling (based on industry experts or market research data, which is fast but unreliable).
Improved user model construction method: user model based on user behavior data
Advantages: simplify the traditional way, reduce the threshold of data analysis; make data analysis more scientific, efficient, and comprehensive, which can be more directly applied to business growth and guide operational strategy.
Methods:
1. Organize and collect initial knowledge of users
2. Segment users
3. Analyze user behavioral data
4. Speculate on target motivations
5. Conduct interviews and surveys with users for validation
6. Modify the establishment of user models
At the same time, it also The collected user information can be mapped into user's attributes or user's behavioral information and stored to form a user profile; real-time attention to the fluctuation of their own data, and make strategic adjustments in a timely manner.
Two, event model
Event model is the first step in the analysis of user behavior data, but also the core and foundation of the analysis, the data structure behind it, the timing of the collection and the management of the event is the three major elements in the event model.
What is an event?
Event is the user's behavior on the product, it is a professional description of the user's behavior, all the user's program feedback obtained on the product can be abstracted as an event, which can be collected by the developers through the buried points. As an example: a user clicking a button on a page is an event.
Event collection
Event-attribute-value structure: event (the user's behavior on the product), attribute (the dimension that describes the event), value (the content of the attribute)
In the process of event collection, the flexible use of the event-attribute-value structure can not only maximize the restoration of the user's use of the scenarios, but also can greatly save the amount of events and improve work efficiency.
The timing of collection: user clicks, web page loading is completed, and the server judges the return. When designing a buried demand document, the timing of collection is particularly important, and is the core of ensuring data accuracy.
An example: the event collection of the e-commerce sales page
Analysis of events
The analysis of the event usually has the number of people triggered by the event, the number of times, the number of times per capita, and the active ratio of the four dimensions of the calculation.
Management of Events
When there are many events, events are grouped and important events are labeled so that they can be managed in different categories. At the same time, important user behavior can be marked out from the product business perspective, so that it is easy and quick to find the use of common and important events in the analysis.
Three, funnel model
Funnel model first originated from the traditional industry marketing business activities evolved, it is a set of process data analysis methods.
The main model framework: by detecting the starting point of the target process (user entry) to the final completion of the target action. The amount of users and retention of each node experienced in this, to assess the good and bad of each node, to find the most need to optimize the node. The funnel model is an important analysis model of the state of user behavior and the conversion rate of users at each stage from the starting point to the end point.
Four, heat map analysis -- drawing user behavior
Heat map, is the most intuitive tool to record user interaction with the product interface. Heatmap analysis, is by recording the user's mouse behavior, and presented in a visual effect, thus helping users to optimize the layout of the site. Whether it is Web or App analysis, heat map analysis is a very important model.
In the actual use of the process, often use several methods of contrast heat map to compare and analyze multiple heat maps to solve the problem:
Comparative analysis of multiple heat maps, in particular, the click heat map (touch heat map), reading heat map, stop-screen heat map;
Comparative analysis of heat maps of subdivided populations, for example, different channels, new and old users, and different time periods, AB test heat map analysis;
Different depth of interaction, the heat map reflected is also different. For example, click heatmap and conversion heatmap comparison analysis;
Five, custom retention analysis
On the concept of retention rate, in the previous article has been introduced. For the product, the higher the retention rate, the more active users of the product, the greater the proportion of loyal users will be converted, the more conducive to the enhancement of the product's cash flow.
Custom retention: Based on the retention of users in their own business scenarios, i.e., the behavior of retention is customized. You can customize the retention behavior by setting the initial behavior and the return visit behavior.
An example: the 5-day retention rate of users who used Harrow*** Enjoy Bike after grabbing the voucher
Initial behavior: grabbing the voucher
Returning behavior: using Harrow*** Enjoy Bike
Sixth, Stickiness Analysis
Stickiness: scientifically assessing the product's ability to stay in the market from users' point of view
Through the stickiness analysis of the users, you can understand how many days in a week or a month the user actually uses your product or even a certain function, and further analyze the user's habit of using the product.
Stickiness analysis is one of the features of Zhuge io, which includes overall product stickiness, functional stickiness, stickiness trend and user group comparison, you can refer to /advanced/stickiness.html
Seven, full behavioral path analysis
The full behavioral path analysis is a unique type of data analysis method for Internet products.
The full behavioral path analysis is a kind of data analysis method unique to Internet products, which analyzes the flow pattern and characteristics of each module in the App or website based on each user's behavioral events in the App or website, and explores the user's access or browsing patterns to achieve some specific business purposes, such as the enhancement of the arrival rate of the core module of the App, the extraction of the mainstream paths of a specific group of users and browsing characterization, and the optimization of the product design of the App.
There are two types of behavioral path models commonly used in the visualization process:
Tree diagram: reflecting the user's behavioral path in a tree structure
Sun diagram: reflecting the user's behavioral path in a ring diagram
In the above diagram, each ring represents one step of the user, and different colors represent different behaviors, and the greater the ratio of the same ring's color, the more unified the user's behavior is in the current step, and the longer the ring is. The larger the proportion of the same ring represents the more uniform user behavior in the current step, and the longer the ring represents the longer the user's behavior path.
Eight, user grouping model
User grouping is the labeling of user information, through the user's historical behavior path, behavioral characteristics, preferences and other attributes, will have the same attributes of the user is divided into a group, and subsequent analysis.
Segmentation model based on user behavioral data: when returning to the behavioral data itself, it will be found that the insight into the user can be more refined and retrospective, and the historical behavioral records can be used to find the desired group of people more quickly.
Four user segmentation dimensions:
User attributes: age, gender, city, browser version, system version, operation version, channel source, etc.;
Active in: by setting the active time, find the active users in the specified interval;
Done/didn't do: analyze the "closeness" of the user's interactions with the product by whether the user performs a certain behavior. "intimacy";
Added in: by setting the time period, the precise screening of the new user's time range;
How to improve product sales is a comprehensive issue, the need to combine a variety of models to analyze the data, the above is a summary of some of the knowledge, and I hope to be able to help you.